Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

Move rows from TableA into Table-Archive

Tags:

mysql

archive

Is it possible to move rows that are 3 days old into an other table called "Table_Archive" automatically in mysql ones a week?

tableA ex:

ID | stringvalue | Timestamp
1  | abc         | 2011-10-01
2  | abc2        | 2011-10-02
3  | abc3        | 2011-10-05
4  | abc4        | 2011-10-10
5  | abc5        | 2011-10-11

After the move

tableA:

ID | stringvalue | Timestamp
4  | abc4        | 2011-10-10
5  | abc5        | 2011-10-11

Table_Archive:

ID | stringvalue | Timestamp
1  | abc         | 2011-10-01
2  | abc2        | 2011-10-02
3  | abc3        | 2011-10-05

And when new input comes into tableA it wont be any problems with ID (PK) in the next move?

What Ive got:

CREATE PROCEDURE clean_tables ()
BEGIN
    BEGIN TRANSACTION;

    DECLARE _now DATETIME;
    SET _now := NOW();

    INSERT
    INTO    Table_Archive
    SELECT  *
    FROM    TableA
    WHERE   timestamp < _now - 3;
    FOR UPDATE;

    DELETE
    FROM    TableA
    WHERE   timestamp < _now - 3;

    COMMIT;
END

How do I change _now to be the date 3 days ago?

like image 423
Linqan Avatar asked Oct 11 '11 10:10

Linqan


1 Answers

Personally, I would make use of the MySQL Event Scheduler. This is a built in event scheduler rather like CRON in Linux.

You can specify it to call a procedure, procedures or functions or run a bit of SQL at designated intervals.

Read the MySQL docs but an example would be:

CREATE EVENT mydatabase.myevent
ON SCHEDULE EVERY 1 WEEK STARTS CURRENT_TIMESTAMP + INTERVAL 10 MINUTE
DO
 call clean_tables();

So this is saying "call clean_tables() once a week and make the first call in 10 minutes' time"

One gotcha is that the event scheduler is (I think) disabled by default. To turn it on run:

SET GLOBAL event_scheduler = ON;

You can then run:

SHOW PROCESSLIST;

To see whether the event scheduler thread is running.

As for preserving your Table A ID column (if you must). I would keep the ID on Table_Archive as unique to that table i.e make it the primary key & auto_increment and then have a 'Original_TableA_ID' column in which to store the TableA ID. You can put a unique index on this if you want.

So Table_Archive would be like:

create table `Table_Archive` (
ID int unsigned primary key auto_increment, -- < primary key auto increment
tableAId unsigned int not null, -- < id column from TableA
stringValue varchar(100),
timestamp datetime,
UNIQUE KEY `archiveUidx1` (`tableAId`) -- < maintain uniqueness of TableA.ID column in Archive table
);

Nobody seems to have answered your original question "How do I change _now to be the date 3 days ago?". You do that using INTERVAL:

DELIMITER $

CREATE PROCEDURE clean_tables ()
BEGIN
BEGIN TRANSACTION;

DECLARE _now DATETIME;
SET _now := NOW();

INSERT
INTO    Table_Archive
SELECT  *
FROM    TableA
WHERE   timestamp < _now - interval 3 day;
FOR UPDATE;

DELETE
FROM    TableA
WHERE   timestamp < _now - interval 3 day;

COMMIT;
END$

DELIMITER ;

One final point is that you should consider creating an index on the timestamp column on TableA to improve the performance of you clean_tables() procedure.

like image 72
Tom Mac Avatar answered Oct 01 '22 08:10

Tom Mac